This work studied the merits of current transfer learning in medical imaging. It represents a strong empirical analysis of current state-of-the-art approaches, and leads to some somewhat surprising conclusions. Overall the reviewers agreed the work was strong and merited accepted.